Possibly the most important questions in AI are not the design of new algorithms, but how to make sure that the data sets with which we train them are not inherently biased. Cathy O’Neil’s book Weapons of Math Destruction is essential reading on this topic.
